You know the combat system is FUBAR when...

[Marsden]

Could there be some kind of intentional randomness introduced into the game so that some strange results are more likely to happen than not? How could you actually play out the battle at Themopolaye in Civ 3? My one elite hopilite vs how many Persian archers, swordsmen and Immortals?
How many units would my Hoplite have to kill before dying? If I was Persia, would I start a thread over it?

 

[Pyrrhos]

I can understand my request for a save this time may have surprised you - however, for future reference, the next time you have incontrovertible proof of AI bias in the RNG, please be aware that I would love to see a copy of the save and would greatly appreciate it if you would post it before deleting it.

I am so tired of excuses

Oh FFS! Cut the ...

These are not the exact but similar combats:

In the first, try to defeat the swordsman, spearman and archer with the slavemaster in one turn without losing any hp. The slavemaster with 12-48-3 should certainly be able to do so on a regular basis as these units are on desert/grassland and not fortified. However, I have yet been unable to do so, although I must confess I only tried some 25 times.

In the second, can the 4-4-1 legion defeat the 1-2-2 Mohican without losing any hp? This far, after 32 attempts, the answer is no although the mohican has slain the legion five times.

Good luck!

 

[Bartleby]

There's probably a big pile of inscribed clay tablets in some cave somewhere.

Edit: crosspost. My weak joke is obviously aimed at Marden's post.

 

[Pyrrhos]

Could there be some kind of intentional randomness introduced into the game so that some strange results are more likely to happen than not? How could you actually play out the battle at Themopolaye in Civ 3? My one elite hopilite vs how many Persian archers, swordsmen and Immortals?
How many units would my Hoplite have to kill before dying? If I was Persia, would I start a thread over it?

Although I appreciate your point, Thermopylae is not a fair comparison as the Greek Alliance defended a pass where the vastly superior Persian army could only bring a small part of its force into contact with the extremely well positioned greeks. After they had been betrayed and the Persians could get at them from the rear, King Leonidas of Sparta dismissed the major part of the army and made the famous last stand with 300 Spartans, 700 Thespian volunteers, 400 Thebans and 900 Helots.

I'd say that a "fair" civ representation would be three numidian mercenaries plus two archers fortified in a fortification on a mountaintop bottleneck being attacked by, roughly, six horsemen + twelve swordsmen + six archers - since Xerxes att no time could bring his whole force to bear even if, in toto, the Greeks of the last stand were outnumbered 100:1 or thereabouts.

In civ, the defensive factors could be represented but not the limitation on how many units could attack per turn.

 

[Chamnix]

These are not the exact but similar combats:


Good luck!

Thank you . I won't have access to civ again for at least 8-12 hours, but I will let you know what I find .

 

[Pyrrhos]

In the second, can the 4-4-1 legion defeat the 1-2-2 Mohican without losing any hp? This far, after 32 attempts, the answer is no although the mohican has slain the legion five times.

I tried this 50 times tonight and wrote down the results of each combat. If you look at the total numbers, the elite legion won 28 and lost three with the veteran mohikan running away 19 times. This looks pretty straightforward.

However, looking closer there are a few hints that all is not right. The legion lost a total of 117 rds against the mohikan's 170. This works out as 2.34 and 3.40 hp lost per combat respectively. i.e. the very much inferior mohikan lost only 1hp more per battle on average. Furthermore, of all the 28 battles the legion won, it lost hp in no less than 25. This is not what you'd expect with the mohikan's attack value of one versus the legion's defence of four.

Furthermore, but still in the realm of the quite possible is that there was no MGL during this run. The chance of one not appearing in 28 won combats with a mil civ is 8.7%. Judging by the number of slaves I have, 166 not counting ex-barbarians, I cannot have won less than 500 combats and have had only four MGLs to show for it. That's way too few.

So yes, something is not quite as it should be here. Wonder how Chamnix is doing and what he finds?

 

[Chamnix]

I have just downloaded the files and haven’t run anything yet, but after reading your post above, I think the problem may just be that you misunderstand the combat system:

I tried this 50 times tonight and wrote down the results of each combat. If you look at the total numbers, the elite legion won 28 and lost three with the veteran mohikan running away 19 times. This looks pretty straightforward.

However, looking closer there are a few hints that all is not right. The legion lost a total of 117 rds against the mohikan's 170. This works out as 2.34 and 3.40 hp lost per combat respectively. i.e. the very much inferior mohikan lost only 1hp more per battle on average. Furthermore, of all the 28 battles the legion won, it lost hp in no less than 25. This is not what you'd expect with the mohikan's attack value of one versus the legion's defence of four.

(bold added)

If the legion attacks the Mohican, the only numbers that are relevant are the legion’s attack value and the Mohican’s defense value; the legion’s defense of 4 and the Mohican’s attack of 1 don’t have any effect whatsoever. The way the combat works is that the legion has an attack strength of 4. The Mohican has a modified defense strength of 2 * (1 + 25%) = 2.5 because he is in a forest.

For a single round of combat, the legion will win 4 ÷ (4+2.5) = 61.54% of the time. The Mohican will win 2.5 ÷ (4+2.5) = 38.46% of the time. Out of 287 rounds as you posted above, you would expect the legion to win 177 times and the Mohican 110 times. Your results of 170-117 are definitely reasonable.

Furthermore, but still in the realm of the quite possible is that there was no MGL during this run. The chance of one not appearing in 28 won combats with a mil civ is 8.7%. Judging by the number of slaves I have, 166 not counting ex-barbarians, I cannot have won less than 500 combats and have had only four MGLs to show for it. That's way too few.

You have an MGL already who has not been used yet. Your chances of getting a new MGL while you already have an unused one are 0%.

Let me know if you still think it is necessary to explore this further.

 

[Theov]

I stopped to complain about it.
I can go very Napoleon on my enemies.

 

[AutomatedTeller]

I see nothing to indicate that the combat system is working differently than described by Firaxis in what has been described.

What you would expect, given about a 61% chance of winning every combat round, is that about 77% of the time, the legion comes through unscratched (.61*.61*.61). What you found is that it happened 89% of the time IN THE CASES WHERE THE IMPI DIDN'T RUN. If it had happened 3 more times (22 out of 28), it would be about right.

If you flipped a coin 50 times and got 22 heads and 28 tails, would you say that the coin is rigged? What if you flipped it 100 times and got 39 heads and 61 tails? Is that a rigged coin?

How many of the times the impi ran away was the legion unscratched?

See, here's the thing - we see the larger number and we think "wow, that should win all the time", and it doesn't mean that, at all. It means an advantage, that's all.

 

[Gosnork]

Judging by the number of slaves I have, 166 not counting ex-barbarians, I cannot have won less than 500 combats and have had only four MGLs to show for it. That's way too few.

I recently had 3 consecutive Elite Tank battles result in MGLs. I had built the Heroic Epic, so the odds of that happening were (1/12)^3, or .05%. It was a Monarch game as the Iroquois and I was the most powerful nation in the world by every possible measurement (luxes, area, technology, military strength). Needless to say, it was a huge challenge to remember to load all those new armies the next turn.

 

[Pyrrhos]

The way the combat works is that the legion has an attack strength of 4. The Mohican has a modified defense strength of 2 * (1 + 25%) = 2.5 because he is in a forest.

For a single round of combat, the legion will win 4 ÷ (4+2.5) = 61.54% of the time. The Mohican will win 2.5 ÷ (4+2.5) = 38.46% of the time.

If it works this way, then it is indeed FUBAR. This means, as one example, that a modern paratrooper (6-11-1) attacking a spearman (1-2-1) on a mountain (def bounus 100%) only has a 6/6+4=60% chance of winning. The Bronze Age spearman has a 40% chance to win...

Another example: A berserk 6-2-1 attacking a modern armour (24-16-3) in the open, let us say on plains, has a 6/6+16=27% chance of winning. If it attacks a tank (16-8-0), the berserk has a 6/6+8=43% chance of victory ie more or less a 50-50 chance.

If this isn't FUBAR...

 

[D0NIMATRIX]

What you mean is that the system itself isn't FUBAR, it's the chances produced by it that are.

 

[Pyrrhos]

You have an MGL already who has not been used yet. Your chances of getting a new MGL while you already have an unused one are 0%.

I recently had 3 consecutive Elite Tank battles result in MGLs.

Hmm...

Not that I have any saves to prove it, but I seem to recall having more than one unconverted MGL at the same time on several occasions. But, I'll take your word for it.

 

[Pyrrhos]

How many of the times the impi ran away was the legion unscratched?

Legion - Mohawk (ie Impi) 0-3r three times
Legion - Mohawk 1-3r five times
Legion - Mohawk 2-3r twice
Legion - Mohawk 3-3r three times
Legion - Mohawk 4-3r twice

It would seem that the state of the opponent is immaterial. The unit with the retreat ability will do so appx 1/3 when only one hp remains.

 

[Gosnork]

If it works this way, then it is indeed FUBAR. This means, as one example, that a modern paratrooper (6-11-1) attacking a spearman (1-2-1) on a mountain (def bounus 100%) only has a 6/6+4=60% chance of winning. The Bronze Age spearman has a 40% chance to win...

Another example: A berserk 6-2-1 attacking a modern armour (24-16-3) in the open, let us say on plains, has a 6/6+16=27% chance of winning. If it attacks a tank (16-8-0), the berserk has a 6/6+8=43% chance of victory ie more or less a 50-50 chance.

If this isn't FUBAR...

How did you think it worked?

Also don't forget that plains offers a 10% bonus to defenders, everything offers at least 10% defensive bonus.

I should have clarified, as soon as my Elite Tanks generated a MGL, I turned it into an Army.

EDIT:

It would seem that the state of the opponent is immaterial. The unit with the retreat ability will do so appx 1/3 when only one hp remains.

In the editor, under the Combat Experience tab, there is a "Retreat Bonus" for each Experience Level. I don't know how this number relates to combat, but you could set it to 0, and see if the Mohawk retreats. Then set it to 100 and see how often it retreats.

Conscript Retreat Bonus: 34
Regular Retreat Bonus: 50
Veteran Retreat Bonus: 58
Elite Retreat Bonus: 66

 

[Pyrrhos]

What you mean is that the system itself isn't FUBAR, it's the chances produced by it that are.

When I said that the combat system is FUBAR I meant so. If the chances produced by it are ridiculous, the system is too. It works, granted, and when Civ I made its debut, it was a good system as most PCs rarely had more than 500kB available for gaming purposes. But it's very disappointing that such a grossly oversimplified system is present in a game from 2001. Panzer General of 1993 provided a much, much better combat system within those limitations. You would have expected that when they jazzed up the graphics and rules otherwise, they might have devoted some attention to the combat system.

 

[D0NIMATRIX]

The combat system does what it has to do: it calculates chances. The game programming does what it has to do with the results: it kills units.

 

[Pyrrhos]

How did you think it worked?

Judging from the graphics/animations and when damage occurrs:

1. Defender free round -> 50% chance to damage x Defender's A/Attacker's D

2. Attacker's round -> Attacker's A + rng vs defender's D + bonuses + rng, the greater score wins the round

3. Defender's A +rng vs Attacker's D + bonuses + rng, the greater score wins the round

4. A

5. D

(etc til one party is "killed" or flees)

Ie. the combat system used in AD&D PC-games and early Military games.

 

[Pyrrhos]

GNBNA, Panzer General, AOD & Civ I - aaah, those were the days!

Interesting info, thx!

 

[timerover51]

I suspect that part of the problem with the games in that they were designed by programmers, not game designers who had worked with miniature or board war games.